Design adaptation methods for genetic association studies

Genetic association studies have become the most widely used gene mapping tool for the identification of disease susceptibility loci of complex common traits or diseases. In order to obtain sufficient power at a certain significance level (type I error risk), these analyses require a complete pre-specification of the total number of individuals to be sampled and genotyped. However, in most of these studies little information about the genetic effect size is available beforehand and thus it is difficult to calculate a reasonable sample size. By addressing these problems, this thesis aims at introducing, extending, and evaluating statistical methodology on design adaptations for genetic association studies. In particular, it is shown how a confirmatory candidate gene association study can be planned and analyzed given the mentioned uncertainties. For this purpose, an adaptive group sequential procedure is developed. If no rejection of the null hypothesis is possible at the interim analysis, the design of the subsequent study part can be modified depending on interim data. As an example, sample size reassessment may be done using interim effect size estimates developed by the author of this thesis, as well. Finally, a new flexible two-stage design for genomewide association studies is presented. While providing strong control of the genomewide, family-wise type I error rate, the new design might also be more cost-efficient due to greater flexibility in comparison to all previously suggested designs. Examples of which are the possibility to base marker selection upon biological criteria instead of statistical criteria or the option to modify the sample size at any time during the course of the project. Both the candidate and the genomewide association designs are evaluated using simulated and real data sets. Finally, prospects and limits of design adaptation methods and estimators of genetic effects in genetic association studies of complex traits are discussed.